Advancements in Structural Biology ­ How to See Molecules of Life? / Postepy biologii strukturalnej ­ jak zobaczyc czasteczki zycia?

Structural biology is focused on understanding the architecture of biomolecules, such as proteins and nucleic acids. Deciphering the structure helps to understand their function in the cell at a very precise ­ molecular level. This makes it possible to not only determine the basis of diseases but also to propose therapeutic strategies and tools. Such a strong motivation for the development of structural biology has led to the development of a number of methods, which enable determination of the structures of the molecules of life. The continuous progress has been enabled by the integration of biology, chemistry, physics, and computer science, making structural biology extremely interdisciplinary. In its 35-year history, the Institute of Bioorganic Chemistry of the Polish Academy of Sciences in Poznan has become one of the key Polish institutions conducting research in the field of structural biology. On one hand, the research has brought international recognition, and on the other hand, it has forced the implementation and development of cutting-edge methods. This review discusses the methods used in structural biology at the Institute.

Structural and biochemical insights into CRISPR RNA processing by the Cas5c ribonuclease SMU1763 from Streptococcus mutans.

The cariogenic pathogen Streptococcus mutans contains two CRISPR systems (type I-C and type II-A) with the Cas5c protein (SmuCas5c) involved in processing of long CRISPR RNA transcripts (pre-crRNA) containing repeats and spacers to mature crRNA guides. In this study, we determined the crystal structure of SmuCas5c at a resolution of 1.72 Å, which revealed the presence of an N-terminal modified RNA recognition motif and a C-terminal twisted ß-sheet domain with four bound sulphate molecules. Analysis of surface charge and residue conservation of the SmuCas5c structure suggested the location of an RNA-binding site in a shallow groove formed by the RNA recognition motif domain with several conserved positively charged residues (Arg39, Lys52, Arg109, Arg127, and Arg134). Purified SmuCas5c exhibited metal-independent ribonuclease activity against single-stranded pre-CRISPR RNAs containing a stem-loop structure with a seven-nucleotide stem and a pentaloop. We found SmuCas5c cleaves substrate RNA within the repeat sequence at a single cleavage site located at the 3'-base of the stem but shows significant tolerance to substrate sequence variations downstream of the cleavage site. Structure-based mutational analysis revealed that the conserved residues Tyr50, Lys120, and His121 comprise the SmuCas5c catalytic residues. In addition, site-directed mutagenesis of positively charged residues Lys52, Arg109, and Arg134 located near the catalytic triad had strong negative effects on the RNase activity of this protein, suggesting that these residues are involved in RNA binding. Taken together, our results reveal functional diversity of Cas5c ribonucleases and provide further insight into the molecular mechanisms of substrate selectivity and activity of these enzymes.

Molecular structure of a U•A-U-rich RNA triple helix with 11 consecutive base triples.

Three-dimensional structures have been solved for several naturally occurring RNA triple helices, although all are limited to six or fewer consecutive base triples, hindering accurate estimation of global and local structural parameters. We present an X-ray crystal structure of a right-handed, U•A-U-rich RNA triple helix with 11 continuous base triples. Due to helical unwinding, the RNA triple helix spans an average of 12 base triples per turn. The double helix portion of the RNA triple helix is more similar to both the helical and base step structural parameters of A'-RNA rather than A-RNA. Its most striking features are its wide and deep major groove, a smaller inclination angle and all three strands favoring a C3'-endo sugar pucker. Despite the presence of a third strand, the diameter of an RNA triple helix remains nearly identical to those of DNA and RNA double helices. Contrary to our previous modeling predictions, this structure demonstrates that an RNA triple helix is not limited in length to six consecutive base triples and that longer RNA triple helices may exist in nature. Our structure provides a starting point to establish structural parameters of the so-called 'ideal' RNA triple helix, analogous to A-RNA and B-DNA double helices.

Base pairing, structural and functional insights into N4-methylcytidine (m4C) and N4,N4-dimethylcytidine (m42C) modified RNA.

The N4-methylation of cytidine (m4C and m42C) in RNA plays important roles in both bacterial and eukaryotic cells. In this work, we synthesized a series of m4C and m42C modified RNA oligonucleotides, conducted their base pairing and bioactivity studies, and solved three new crystal structures of the RNA duplexes containing these two modifications. Our thermostability and X-ray crystallography studies, together with the molecular dynamic simulation studies, demonstrated that m4C retains a regular C:G base pairing pattern in RNA duplex and has a relatively small effect on its base pairing stability and specificity. By contrast, the m42C modification disrupts the C:G pair and significantly decreases the duplex stability through a conformational shift of native Watson-Crick pair to a wobble-like pattern with the formation of two hydrogen bonds. This double-methylated m42C also results in the loss of base pairing discrimination between C:G and other mismatched pairs like C:A, C:T and C:C. The biochemical investigation of these two modified residues in the reverse transcription model shows that both mono- or di-methylated cytosine bases could specify the C:T pair and induce the G to T mutation using HIV-1 RT. In the presence of other reverse transcriptases with higher fidelity like AMV-RT, the methylation could either retain the normal nucleotide incorporation or completely inhibit the DNA synthesis. These results indicate the methylation at N4-position of cytidine is a molecular mechanism to fine tune base pairing specificity and affect the coding efficiency and fidelity during gene replication.

Peculiar substrate specificity of δ1-pyrroline-5-carboxylate reductase in the obligately fermentative bacterium Zymomonas mobilis.

BACKGROUND: The enzyme that catalyzes the last step in proline synthesis, δ1-pyrroline-5-carboxylate reductase, showed in most cases a distinct preference in vitro for NADPH as the electron donor. METHODS AND RESULTS: A Zymomonas mobilis gene coding for a δ1-pyrroline-5-carboxylate reductase was cloned and heterologously expressed, and the recombinant protein was purified and characterized. The enzyme showed higher affinity to, and higher catalytic rate with NADH, with a specific activity of about 600 nkat (mg protein)-1. The molecular basis of this feature was investigated by analysis of the dinucleotide binding domain in silico. CONCLUSIONS: We postulate that the main determinants of coenzyme preference for P5C reductases are the length and the sequence of the motif A, whereas the overall sequence identity is insufficient to predict it a priori. Results are discussed in view of the obligately fermentative metabolism of this bacterium.

Crystal structures of plant inorganic pyrophosphatase, an enzyme with a moonlighting autoproteolytic activity.

Inorganic pyrophosphatases (PPases, EC 3.6.1.1), which hydrolyze inorganic pyrophosphate to phosphate in the presence of divalent metal cations, play a key role in maintaining phosphorus homeostasis in cells. DNA coding inorganic pyrophosphatases from Arabidopsis thaliana (AtPPA1) and Medicago truncatula (MtPPA1) were cloned into a bacterial expression vector and the proteins were produced in Escherichia coli cells and crystallized. In terms of their subunit fold, AtPPA1 and MtPPA1 are reminiscent of other members of Family I soluble pyrophosphatases from bacteria and yeast. Like their bacterial orthologs, both plant PPases form hexamers, as confirmed in solution by multi-angle light scattering and size-exclusion chromatography. This is in contrast with the fungal counterparts, which are dimeric. Unexpectedly, the crystallized AtPPA1 and MtPPA1 proteins lack ∼30 amino acid residues at their N-termini, as independently confirmed by chemical sequencing. In vitro, self-cleavage of the recombinant proteins is observed after prolonged storage or during crystallization. The cleaved fragment corresponds to a putative signal peptide of mitochondrial targeting, with a predicted cleavage site at Val31-Ala32. Site-directed mutagenesis shows that mutations of the key active site Asp residues dramatically reduce the cleavage rate, which suggests a moonlighting proteolytic activity. Moreover, the discovery of autoproteolytic cleavage of a mitochondrial targeting peptide would change our perception of this signaling process.

Guarding the gateway to histidine biosynthesis in plants: Medicago truncatula ATP-phosphoribosyltransferase in relaxed and tense states.

In the first committed step of histidine biosynthesis, adenosine 5'-triphosphate (ATP) and 5-phosphoribosyl-α1-pyrophosphate (PRPP), in the presence of ATP phosphoribosyltransferase (ATP-PRT, EC 2.4.2.17), yield phosphoribosyl-ATP. ATP-PRTs are subject to feedback inhibition by histidine that allosterically binds between the regulatory domains. Histidine biosynthetic pathways of bacteria, lower eukaryotes, and plants are considered promising targets for the design of antibiotics, antifungal agents, and herbicides because higher organisms are histidine heterotrophs. Plant ATP-PRTs are similar to one of the two types of their bacterial counterparts, the long-type ATP-PRTs. A biochemical and structural study of ATP-PRT from the model legume plant, Medicago truncatula (MedtrATP-PRT1) is reported herein. Two crystal structures, presenting homohexameric MedtrATP-PRT1 in its relaxed (R-) and histidine-bound, tense (T-) states allowed to observe key features of the enzyme and provided the first structural insights into an ATP-PRT from a eukaryotic organism. In particular, they show pronounced conformational reorganizations during R-state to T-state transition that involves substantial movements of domains. This rearrangement requires a trans- to cis- switch of a peptide backbone within the hinge region of MedtrATP-PRT1. A C-terminal α-helix, absent in bacteria, reinforces the hinge that is constituted by two peptide strands. As a result, conformations of the R- and T-states are significantly different from the corresponding states of prokaryotic enzymes with known 3-D structures. Finally, adenosine 5'-monophosphate (AMP) bound at the active site is consistent with a competitive (and synergistic with histidine) nature of AMP inhibition.

Structural Studies of Medicago truncatula Histidinol Phosphate Phosphatase from Inositol Monophosphatase Superfamily Reveal Details of Penultimate Step of Histidine Biosynthesis in Plants.

The penultimate enzyme in the histidine biosynthetic pathway catalyzes dephosphorylation of l-histidinol 1-phosphate (HOLP) into l-histidinol. The recently discovered in Arabidopsis thaliana plant-type histidinol phosphate phosphatase (HPP) shares no homology with the two other HPP superfamilies known previously in prokaryotes and resembles myo-inositol monophosphatases (IMPases). In this work, identification of an HPP enzyme from a model legume, Medicago truncatula (MtHPP) was based on the highest sequence identity to A. thaliana enzyme. Biochemical assays confirmed that MtHPP was able to cleave inorganic phosphate from HOLP but not from d-myo-inositol-1-phosphate, the main substrate of IMPases. Dimers of MtHPP, determined by size exclusion chromatography, in the presence of CO2 or formaldehyde form mutual, methylene-bridged cross-links between Lys(158) and Cys(245) residues. Four high resolution crystal structures, namely complexes with HOLP (substrate), l-histidinol (product), and PO4 (3-) (by-product) as well as the structure showing the cross-linking between two MtHPP molecules, provide detailed structural information on the enzyme. Based on the crystal structures, the enzymatic reaction mechanism of IMPases is accustomed to fit the data for MtHPP. The enzymatic reaction, which requires Mg(2+) cations, is catalyzed mainly by amino acid residues from the N-terminal domain. The C-terminal domain, sharing little identity with IMPases, is responsible for the substrate specificity (i.e. allows the enzyme to distinguish between HOLP and d-myo-inositol-1-phosphate). Structural features, mainly the presence of a conserved Asp(246), allow MtHPP to bind HOLP specifically.

Specific binding of gibberellic acid by cytokinin-specific binding proteins: a new aspect of plant hormone-binding proteins with the PR-10 fold.

Pathogenesis-related proteins of class 10 (PR-10) are a family of plant proteins with the same fold characterized by a large hydrophobic cavity that allows them to bind various ligands, such as phytohormones. A subfamily with only ~20% sequence identity but with a conserved canonical PR-10 fold have previously been recognized as Cytokinin-Specific Binding Proteins (CSBPs), although structurally the binding mode of trans-zeatin (a cytokinin phytohormone) was found to be quite diversified. Here, it is shown that two CSBP orthologues from Medicago truncatula and Vigna radiata bind gibberellic acid (GA3), which is an entirely different phytohormone, in a conserved and highly specific manner. In both cases a single GA3 molecule is found in the internal cavity of the protein. The structural data derived from high-resolution crystal structures are corroborated by isothermal titration calorimetry (ITC), which reveals a much stronger interaction with GA3 than with trans-zeatin and pH dependence of the binding profile. As a conclusion, it is postulated that the CSBP subfamily of plant PR-10 proteins should be more properly linked with general phytohormone-binding properties and termed phytohormone-binding proteins (PhBP).

Targeting imidazole-glycerol phosphate dehydratase in plants: novel approach for structural and functional studies, and inhibitor blueprinting.

The histidine biosynthetic pathway (HBP) is targeted for herbicide design with preliminary success only regarding imidazole-glycerol phosphate dehydratase (IGPD, EC 4.2.1.19), or HISN5, as referred to in plants. HISN5 catalyzes the sixth step of the HBP, in which imidazole-glycerol phosphate (IGP) is dehydrated to imidazole-acetol phosphate. In this work, we present high-resolution cryoEM and crystal structures of Medicago truncatula HISN5 (MtHISN5) in complexes with an inactive IGP diastereoisomer and with various other ligands. MtHISN5 can serve as a new model for plant HISN5 structural studies, as it enables resolving protein-ligand interactions at high (2.2 Å) resolution using cryoEM. We identified ligand-binding hotspots and characterized the features of plant HISN5 enzymes in the context of the HISN5-targeted inhibitor design. Virtual screening performed against millions of small molecules not only revealed candidate molecules but also identified linkers for fragments that were experimentally confirmed to bind. Based on experimental and computational approaches, this study provides guidelines for designing symmetric HISN5 inhibitors that can reach two neighboring active sites. Finally, we conducted analyses of sequence similarity networks revealing that plant HISN5 enzymes derive from cyanobacteria. We also adopted a new approach to measure MtHISN5 enzymatic activity using isothermal titration calorimetry and enzymatically synthesized IGP.

Probing the active site of Class 3 L-asparaginase by mutagenesis. I. Tinkering with the zinc coordination site of ReAV.

ReAV, the inducible Class-3 L-asparaginase from the nitrogen-fixing symbiotic bacterium Rhizobium etli, is an interesting candidate for optimizing its enzymatic potential for antileukemic applications. Since it has no structural similarity to known enzymes with this activity, it may offer completely new ways of approach. Also, as an unrelated protein, it would evade the immunological response elicited by other asparaginases. The crystal structure of ReAV revealed a uniquely assembled protein homodimer with a highly specific C135/K138/C189 zinc binding site in each subunit. It was also shown before that the Zn2+ cation at low and optimal concentration boosts the ReAV activity and improves substrate specificity, which indicates its role in substrate recognition. However, the detailed catalytic mechanism of ReAV is still unknown. In this work, we have applied site-directed mutagenesis coupled with enzymatic assays and X-ray structural analysis to elucidate the role of the residues in the zinc coordination sphere in catalysis. Almost all of the seven ReAV muteins created in this campaign lost the ability to hydrolyze L-asparagine, confirming our predictions about the significance of the selected residues in substrate hydrolysis. We were able to crystallize five of the ReAV mutants and solve their crystal structures, revealing some intriguing changes in the active site area as a result of the mutations. With alanine substitutions of Cys135 or Cys189, the zinc coordination site fell apart and the mutants were unable to bind the Zn2+ cation. Moreover, the absence of Lys138 induced atomic shifts and conformational changes of the neighboring residues from two active-site Ser-Lys tandems. Ser48 from one of the tandems, which is hypothesized to be the catalytic nucleophile, usually changes its hydration pattern in response to the mutations. Taken together, the results provide many useful clues about the catalytic mechanism of the enzyme, allowing one to cautiously postulate a possible enzymatic scenario.

Structural and functional studies of Arabidopsis thaliana glutamate dehydrogenase isoform 2 demonstrate enzyme dynamics and identify its calcium binding site.

Glutamate dehydrogenase (GDH) is an enzyme at the crossroad of plant nitrogen and carbon metabolism. GDH catalyzes the conversion of 2-oxoglutarate into glutamate (2OG → Glu), utilizing ammonia as cosubstrate and NADH as coenzyme. The GDH reaction is reversible, meaning that the NAD+-dependent reaction (Glu → 2OG) releases ammonia. In Arabidopsis thaliana, three GDH isoforms exist, AtGDH1, AtGDH2, and AtGDH3. The subject of this work is AtGDH2. Previous reports have suggested that enzymes homologous to AtGDH2 contain a calcium-binding EF-hand motif located in the coenzyme binding domain. Here, we show that while AtGDH2 indeed does bind calcium, the binding occurs elsewhere and the region predicted to be the EF-hand motif has a completely different structure. As the true calcium binding site is > 20 Å away from the active site, it seems to play a structural, rather than catalytic role. We also performed comparative kinetic characterization of AtGDH1 and AtGDH2 using spectroscopic methods and isothermal titration calorimetry, to note that the isoenzymes generally exhibit similar behavior, with calcium having only a minor effect. However, the spatial and temporal changes in the gene expression profiles of the three AtGDH genes point to AtGDH2 as the most prevalent isoform.

Insights into the substrate specificity, structure, and dynamics of plant histidinol-phosphate aminotransferase (HISN6).

Histidinol-phosphate aminotransferase is the sixth protein (hence HISN6) in the histidine biosynthetic pathway in plants. HISN6 is a pyridoxal 5'-phosphate (PLP)-dependent enzyme that catalyzes the reversible conversion of imidazole acetol phosphate into L-histidinol phosphate (HOLP). Here, we show that plant HISN6 enzymes are closely related to the orthologs from Chloroflexota. The studied example, HISN6 from Medicago truncatula (MtHISN6), exhibits a surprisingly high affinity for HOLP, which is much higher than reported for bacterial homologs. Moreover, unlike the latter, MtHISN6 does not transaminate phenylalanine. High-resolution crystal structures of MtHISN6 in the open and closed states, as well as the complex with HOLP and the apo structure without PLP, bring new insights into the enzyme dynamics, pointing at a particular role of a string-like fragment that oscillates near the active site and participates in the HOLP binding. When MtHISN6 is compared to bacterial orthologs with known structures, significant differences arise in or near the string region. The high affinity of MtHISN6 appears linked to the particularly tight active site cavity. Finally, a virtual screening against a library of over 1.3 mln compounds revealed three sites in the MtHISN6 structure with the potential to bind small molecules. Such compounds could be developed into herbicides inhibiting plant HISN6 enzymes absent in animals, which makes them a potential target for weed control agents.

Impaired interactions of ataxin-3 with protein complexes reveals their specific structure and functions in SCA3 Ki150 model.

Spinocerebellar ataxia type 3 (SCA3/MJD) is a neurodegenerative disease caused by CAG expansion in mutant ATXN3 gene. The resulting PolyQ tract in mutant ataxin-3 protein is toxic to neurons and currently no effective treatment exists. Function of both normal and mutant ataxin-3 is pleiotropic by their interactions and the influence on protein level. Our new preclinical Ki150 model with over 150 CAG/Q in ataxin-3 has robust aggregates indicating the presence of a process that enhances the interaction between proteins. Interactions in large complexes may resemble the real-life inclusion interactions and was never examined before for mutant and normal ataxin-3 and in homozygous mouse model with long polyQ tract. We fractionated ataxin-3-positive large complexes and independently we pulled-down ataxin-3 from brain lysates, and both were followed by proteomics. Among others, mutant ataxin-3 abnormally interacted with subunits of large complexes such as Cct5 and 6, Tcp1, and Camk2a and Camk2b. Surprisingly, the complexes exhibit circular molecular structure which may be linked to the process of aggregates formation where annular aggregates are intermediate stage to fibrils which may indicate novel ataxin-3 mode of interactions. The protein complexes were involved in transport of mitochondria in axons which was confirmed by altered motility of mitochondria along SCA3 Ki150 neurites.

The effects of nature-inspired amino acid substitutions on structural and biochemical properties of the E. coli L-asparaginase EcAIII.

The Escherichia coli enzyme EcAIII catalyzes the hydrolysis of L-Asn to L-Asp and ammonia. Using a nature-inspired mutagenesis approach, we designed and produced five new EcAIII variants (M200I, M200L, M200K, M200T, M200W). The modified proteins were characterized by spectroscopic and crystallographic methods. All new variants were enzymatically active, confirming that the applied mutagenesis procedure has been successful. The determined crystal structures revealed new conformational states of the EcAIII molecule carrying the M200W mutation and allowed a high-resolution observation of an acyl-enzyme intermediate with the M200L mutant. In addition, we performed structure prediction, substrate docking, and molecular dynamics simulations for 25 selected bacterial orthologs of EcAIII, to gain insights into how mutations at the M200 residue affect the active site and substrate binding mode. This comprehensive strategy, including both experimental and computational methods, can be used to guide further enzyme engineering and can be applied to the study of other proteins of medicinal or biotechnological importance.

Deciphering the structure of Arabidopsis thaliana 5-enol-pyruvyl-shikimate-3-phosphate synthase: An essential step toward the discovery of novel inhibitors to supersede glyphosate.

Glyphosate interferes with plant aromatic metabolism through the inhibition of 5-enol-pyruvyl-shikimate-3-phosphate (EPSP) synthase [EPSPS, EC 2.5.1.19]. For this reason, EPSPS has been extensively studied in a vast array of organisms. This notwithstanding, up to date, the crystal structure of the protein has been solved exclusively in a few prokaryotes, while that of the plant enzyme has been only deduced in silico by similarity. This study aimed at determining the structure of EPSPS from the plant model species Arabidopsis thaliana, which has been cloned, heterologously expressed and affinity-purified. The kinetic properties of the enzyme have been determined, as well as its susceptibility to the inhibition brought about by glyphosate. The crystal structure of the protein has been resolved at high resolution (1.4 Å), showing open conformation of the enzyme, which is the state ready for substrate/inhibitor binding. This provides a framework for the structure-based design of novel EPSPS inhibitors. Surface regions near the active-site cleft entrance or at the interdomain hinge appear promising for inhibitor selectivity, while bound chloride near the active site is a potential placeholder for anionic moieties of future herbicides.

Functional Characterization of Saccharomyces cerevisiae P5C Reductase, the Enzyme at the Converging Point of Proline and Arginine Metabolism.

The enzyme that, in Saccharomyces cerevisiae, catalyzes the last step in both proline synthesis and arginine catabolism, δ1-pyrroline-5-carboxylate (P5C) reductase, was purified to near homogeneity and characterized thoroughly. Retention patterns upon gel permeation chromatography were consistent with a homodecameric composition of the holomer. High lability of the purified preparations and stabilization by reducing compounds suggested susceptibility to reactive-oxygen-species-mediated damage. Both NADH and NADPH were used as the electron donor, the latter resulting in a 3-fold higher Vmax. However, a higher affinity toward NADH was evident, and the NADPH-dependent activity was inhibited by NAD+, proline, arginine, and a variety of anions. With proline and arginine, the inhibition was of the competitive type with respect to the specific substrate, and of the uncompetitive- or mixed-type with respect to NADPH, respectively. The results suggest that, contrary to the enzyme from higher plants, yeast P5C reductase may preferentially use NADH in vivo. An in silico analysis was also performed to investigate the structural basis of such enzyme features. Superposition of the protein model with the experimental structure of P5C reductase from Medicago truncatula allowed us to hypothesize on the possible allosteric sites for arginine and anion binding, and the cysteine pairs that may be involved in disulfide formation.

Arabidopsis thaliana serine hydroxymethyltransferases: functions, structures, and perspectives.

Serine hydroxymethyltransferase (SHM) is one of the hallmarks of one-carbon metabolism. In plants, isoforms of SHM participate in photorespiration and/or transfer the one-carbon unit from L-serine to tetrahydrofolate (THF), hence producing 5,10-CH2-THF that is needed, e.g., for biosynthesis of methionine, thymidylate, and purines. These links highlight the importance of SHM activity in DNA biogenesis, its epigenetic methylations, and in stress responses. Plant genomes encode several SHM isoforms that localize to cytosol, mitochondria, plastids, and nucleus. In this work, we present a thorough functional and structural characterization of all seven SHM isoforms from Arabidopsis thaliana (AtSHM1-7). In particular, we analyzed tissue-specific expression profiles of the AtSHM genes. We also compared catalytic properties of the active AtSHM1-4 in terms of catalytic efficiency in both directions and inhibition by the THF substrate. Despite numerous attempts to rescue the SHM activity of AtSHM5-7, we failed, which points towards different physiological functions of these isoforms. Comparative analysis of experimental and predicted three-dimensional structures of AtSHM1-7 proteins indicated differences in regions that surround the entrance to the active site cavity.

New aspects of DNA recognition by group II WRKY transcription factor revealed by structural and functional study of AtWRKY18 DNA binding domain.

WRKY transcription factors (TFs) constitute one of the largest families of plant TFs. Based on the organization of domains and motifs, WRKY TFs are divided into three Groups (I-III). The WRKY subgroup IIa includes three representatives in A. thaliana, AtWRKY18, AtWRKY40, and AtWRKY60, that participate in biotic and abiotic stress responses. Here we present crystal structures of the DNA binding domain (DBD) of AtWRKY18 alone and in the complex with a DNA duplex containing the WRKY-recognition sequence, W-box. Subgroup IIa WRKY TFs are known to form homo and heterodimers. Our data suggest that the dimerization interface of the full-length AtWRKY18 involves contacts between the DBD subunits. DNA binding experiments and structural analysis point out novel aspects of DNA recognition by WRKY TFs. In particular, AtWRKY18-DBD preferentially binds an overlapping tandem of W-boxes accompanied by a quasi-W-box motif. The binding of DNA deforms the B-type double helix, which suggests that the DNA fragment must be prone to form a specific structure. This can explain why despite the short W-box consensus, WRKY TFs can precisely control gene expression. Finally, this first experimental structure of a Group II WRKY TF allowed us to compare Group I-III representatives.